Artificial Intelligent Systems and Machine Learning

R. Indra Gandhi ¹, K. Iyakutti ²

Affiliations
1 Department of Computer Science, Mother Teresa Women's University, Kodaikanal, Tamil Nadu, IN
2 School of Physics, Madurai Kamaraj University, Madurai, Tamil Nadu, IN

Abstract

Segmentation is an important phase towards the designing of optical character recognition system. One of the important reasons for poor recognition rate in OCR system is due to incorrect segmentation of characters. Most of the segmentation algorithms primarily aim at segmenting text, graphics, page, line and word. Character segmentation is the fundamental process in character recognition approaches, which rely on isolated characters. Sometimes during segmentation, characters of same word touch each other thus producing vertically overlapping characters. In Tamil scripts, applying the simple concept of vertical projection in segmenting the whole document into individual characters does not work well. As a first step in resolving this, this paper presents an intelligent technique for solving the key problems of distorted merging (touching) characters segmentation. The results show that the proposed algorithm yields promising segmentation output and feasible with other existing techniques, easy for extension, and may be very effective for non-headline based complex Indic scripts.

Keywords

Segmentation, Distorted Character Segmentation, Vertical Overlapping, Merging (Touching) Character Segmentation, Non-Headline Scripts.

Full Text